Spinoff2: Using orientation to control position 1

[fsincox]

The whole idea of using orientation to control location has also come up in my past work as the “GD&T guy”. A very well respected senior inspector proposed the use of parallelism to control the location of shaft keyways. His idea was that you would line up the shaft diameter (the primary datum) and then roll the keyway, itself (the secondary datum) to center and proceed to measure the resulting parallelism of the opposed sides of the keyway. His rational was that the resulting parallelism of the sides of the keyway in this representation of the installed state was important to the key not working its way out under heavy loading. I have to admit I did not feel it was technically illegal, what do you think?

 

[pmarc]

Okay, I am out of bullets. But at least my reply is #100.

 

[Belanger]

Dean, I don't have the standard in front of me, but does it really say that the datum feature simulators must be perfectly located to one another?

We all know about the "mutually perpendicular" requirement, but that's perfect orientation. I'm itching to see if it says perfect location.

John-Paul Belanger
Certified Sr. GD&T Professional
Geometric Learning Systems

http://www.gdtseminars.com

 

[fsincox]

I thank you all again for going through this.
This was 20 years ago, technically, we should be asking what it would mean under the 1982 standard. Any part drawing that I am aware of that used this scheme would have been referencing that.

Dean,
Thank you.

CH,
Is pmarcs’ issue the same problem you had too when you proposed A-B, I never really understood where you were going there?

pmarc,
I think your question is now whether the “theoretical gages” we conceptually use to establish these things are allowed to be perfect or based on an “image of perfection”. I admit, I have always thought this was an oversimplification by the standard too, but I believe it is one ASME has been making since the beginning.
This “image of perfection” gaging concept was the way it explained in 1982 and if 1994 has made statements that undercut that then, for me, it will serve as another example of the standards being inconsistent with their own logic. Then if, 2009 makes statements to bring it back, it sounds more to me like someone made a temporary mistake. What was the actual concentricity between the (2) A-B collets used to inspect runout? What is the tolerance on the 25mm basic offset datum A established under the offset datum section? What is the tolerance of datum targets relative to each other? What is the tolerance on a basic dimension in the first place? These examples and concepts are based on the same ‘image of perfection” and have been shown consistently throughout the different versions of the standard all with small text caveats saying in essence: “if it matters, it might matter”. Kind of like the ISO saying: “if the part is out of tolerance but it works the part is OK”, seems like they are all “hedging their bets” to me.
Frank

 

[axym]

J-P,

It really does say that. The simulators must have basic location, at least in 2009. I don't have the standard in front of me either, but I referenced the section earlier in this thread. See my post from Feb 5 at 10:39, Section 4.5.2 c).

Evan Janeshewski

Axymetrix Quality Engineering Inc.

http://www.axymetrix.ca

 

[Belanger]

Doh, thanks Evan. Sounds like case closed (presuming the appropriate standard is referenced). Sorry I doubted you, Dean

John-Paul Belanger
Certified Sr. GD&T Professional
Geometric Learning Systems

http://www.gdtseminars.com

 

[CheckerHater]

fsincox,
About A-B.
I cannot make a good picture right now, so I'll try to explain it verbally. Please try to follow.
Datum A derived from Diameter is an axis, just a line in space; you cannot locate your keyway from just one line, it's against basic mathematics, no matter front door or back door.
Datum B is plane, derived from keyway, floating somewhere in space. You guys just proved brilliantly that nobody knows for sure how A is related to B. Lets agree that even if datums have to be located perfectly, datum features have not, until you explicitly specify relation thru FCF.
Datum A-B in my understanding is plane, located as close as possible to BOTH centroid derived from diameter and centroid derived from keyway. There is only one plane like that, even if it's not perfectly located in space.
Now when we try to control sides of the keyway to plane A-B thru Parallelism or even Angularity, the tighter is the tolerance the closer we are to perfect keyway.
I think this discussion was already long enough, so I kind of slowed down. It is not like I lost interest.

 

[fsincox]

CH,
Are you opposed to more? I actually did have a point to all this and was debating whether it should be a new post or continue. It actually will be along the same concept you proposed in a way. I see the shaft and key as more of a primary-secondary kind of relationship. After the discussion I realized that you may have thought that the use of coequal datums was a way to get around this whole, “but B is not located to A” bit. How can they argue with A-B it is done all of the time, I agree but it whit the argument, but, it does not agree with actual function IMHO.
I really am not trying to tear everything down and start all over as some may like to think. I am just trying to get them to be consistent and live by their own rules, it is supposed to be a standard after all.
Some think standard means force all to do it this way, I think standard means provide a consistent set of tools to use to get any needed job done.
I would like to find a way to contact you if we could, not sure how to do that though?
Frank

 

[CheckerHater]

Thank you Frank,
I was under impression the thread is coming to logical conclusion.
Most people agreed that under some conditions Orientation may serve "quasi-positional" purpose.
You mentioned yourself that Position would be your first choice, so I was seeing the whole thing as secondary "back-up plan" rather than mainstream solution. Maybe I was wrong.
In my experience in today's world even simple textbook solutions (not to mention more "exotic" approaches) have to be shown down one's throat, so keeping things simple should be priority.
Nevertheless I will be happy to join "spin-off" dedicated to any of 15 (or whatever) GD&T symbols.

 

[axym]

I'm not done talking about this. The horse is not quite dead yet ;^). I think there is something in the peculiar geometry of the shaft and keyway.

I was trying to think of other applications in which the "back-door location" technique would be possible. One of them was controlling a cross hole to be centered on the axis of a shaft. Another was controlling a cylindrical hole to be centered on a spherical feature.

In all of these applications, we're trying to "aim" something at a target that is in the middle of something round. The interesting thing (to me) is that the distinction between orientation error and location error becomes blurred. If I'm shooting at a target and I miss, I can describe the problem in two ways:

1) I aimed in the wrong direction for where I was standing. I can correct the problem by standing in the same place but aiming at a different angle.

2) I was standing in the wrong place for the direction I was aiming. I can correct the problem by aiming in the same direction but moving to a different place.

Description 1) amounts to orientation error, and description 2) amounts to location error. It's two different ways of describing the same error, so we could control it in two different ways.

Does that make any sense?

Evan Janeshewski

Axymetrix Quality Engineering Inc.

http://www.axymetrix.ca

 

[CheckerHater]

Evan,
Just imagine flat plate with several holes all pointing to one single "shooting point" in space.
It's either that or simply a matter of different coordinate system, but then it still a "position".

 

[fsincox]

Evan and CH,
Basically I think the same effect that we have been discussing here, is the same effect that makes runout work. Yes, the runout I find somewhat useless as a concept and is supported by the ASME most likely only because it is a popular inspection method.(sorry guys, particularly Dingy3, I know he loves it so). I am not really trying to take his favorite method of inspection away, though some who may want to simplify might.
John Liggett proposed a simplification of GD&T in a Machine Design (Aug 20, 1992) article, in it he refers to runout as an orientation tolerance. Early ASME versions placed runout under tolerances of “form and orientation”. Taking an indicator and running it across a surface is normally an orientation (parallelism) check, it is the particular special datum set up and rotation that forces location control. I suspect that is related to the effect going on here. In essence the particular datum framework establishment controls location with an orientation tolerance!
So if you will insist it is OK to not control the axis in runout why do you want to worry about it on the slot? Let’s be consistent at least, I think the same uncomfortable feeling you get about the slot not being located is exactly the same feeling the ISO and I feel about runout vs. concentricity!
Frank

 

[DeanD3W]

CheckerHater,
For your datum A-B description, the datum feature simulators for A and B are perfect, or built with tooling tolerance, so relative to the part, they can be considered perfect. All the same for A as primary and B as secondary... Datum A is the axis of the datum feature A simulator and datum B is the center plane of the datum feature B simulator. Datum A-B would be a line on a plane, with the only degree of freedom remaining being translation along the line (see Y14.5-2009 Figure 4-3(f)). The line would be the same as datum axis A and the plane would be the same as datum axis A.

All the difference between [A,B] and [A-B] comes when the datum features are mated with their simulators... For the [A,B] case the simulator for A first grows material until it constrains datum feature A as fully as a perfect cylinder can, then the simulator for B grows material (using Regardless of Material Boundary, as with A) to constrain rotation about datum A. For [A-B] the two simulators grow material simultaneously and each will continue to shrink in size until they constrain the datum features as much as they "do"... I say "do" because there is currently a discussion within and between the Y14.5 and Y14.5.1 committees to resolve whether the second simulator will continue to shrink around its datum feature after the other has engaged the datum feature to the point that it cannot shrink any more. With any "A-B" type of case with features of size this is a point that needs clarification in the standard.

With A-B the slot has an equal influence to that of the cylindrical feature for constraining one of the rotational degrees of freedom that the cylinder as a primary would have first influence over... This makes A-B pretty difficult to handle correctly for most inspection methods that I am aware of.

I wouldn't recommend A-B for this type of part. For this discussion and to better model the most common functional cases for this type of part, I think A as primary and B as secondary is better.

Dean

http://www.d3w-engineering.com

 

[CheckerHater]

Dean,
OK, [A] simulator is a chuck or collet.
How you visualize simulator and its relation to A?

 

[DeanD3W]

CH,
The B simulator would be a "double acting vice" (possibly not precise enough example at

http://www.grizzly.com/products/Precision-Self-Centering-Vise/H7576

).

It needs to remain centered such that its center plane and datum A, so the collet's axis, are coincident.

Dean

http://www.d3w-engineering.com